Hindawi Journal of Healthcare Engineering Volume 2017, Article ID 5740975, 10 pages https://doi.org/10.1155/2017/5740975

Research Article Expression Changes in Long-Term In Vitro Human Blood-Brain Barrier Models and Their Dependence on a Transwell Scaffold Material

1 1 2 3 Joel D. Gaston, Lauren L. Bischel, Lisa A. Fitzgerald, Kathleen D. Cusick, 4 2 Bradley R. Ringeisen, and Russell K. Pirlo

1American Society for Engineering Education Postdoctoral Fellowship Program, U.S. Naval Research Laboratory, Washington, DC, USA 2Chemistry Division, U.S. Naval Research Laboratory, Washington, DC, USA 3National Research Council Postdoctoral Fellowship Program, U.S. Naval Research Laboratory, Washington, DC, USA 4Defense Advanced Research Projects Agency, Arlington, VA, USA

Correspondence should be addressed to Russell K. Pirlo; [email protected]

Received 20 June 2017; Revised 19 September 2017; Accepted 12 October 2017; Published 29 November 2017

Academic Editor: George Stan

Copyright © 2017 Joel D. Gaston et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Disruption of the blood-brain barrier (BBB) is the hallmark of many neurovascular disorders, making it a critically important focus for therapeutic options. However, testing the effects of either drugs or pathological agents is difficult due to the potentially damaging consequences of altering the normal brain microenvironment. Recently, in vitro coculture tissue models have been developed as an alternative to animal testing. Despite low cost, these platforms use synthetic scaffolds which prevent normal barrier architecture, cellular crosstalk, and tissue remodeling. We created a biodegradable electrospun gelatin mat “biopaper” (BP) as a scaffold material for an endothelial/astrocyte coculture model allowing cell-cell contact and crosstalk. To compare the BP and traditional models, we investigated the expression of 27 involved in BBB permeability, cellular function, and endothelial junctions at different time points. Gene expression levels demonstrated higher expression of transcripts involved in endothelial junction formation, including TJP2 and CDH5, in the BP model. The traditional model had higher expression of genes associated with extracellular matrix-associated , including SPARC and COL4A1. Overall, the results demonstrate that the BP coculture model is more representative of a healthy BBB state, though both models have advantages that may be useful in disease modeling.

1. Introduction considerable physical space, and often do no not translate well to human results [2, 3]. One way to overcome the short- The unique characteristics of the blood-brain barrier (BBB) comings of in vivo systems is through in vitro models capable present a considerable challenge for studying central nervous of accurately mimicking human systems. In vitro model sys- system (CNS) therapeutics and disease modeling. Lipid- tems consisting of cell-based arrays are a valuable tool to soluble molecules and gases readily diffuse across the BBB screen molecular BBB permeability prior to animal testing. while larger, hydrophilic molecules are impeded by endothe- The most common platform for in vitro BBB testing lial tight junctions [1]. The intrinsic difficulty of crossing the incorporates multiple cell types cultured on transwell inserts, BBB necessitates extensive testing for new drug candidates allowing for separate culture compartments within the same using relevant in vitro and in vivo models. In vivo model well [1]. The transwell inserts consist of a semiporous poly- organisms, such as rats, provide valuable information in drug carbonate or polyethylene terephthalate (PET) membrane. screening assays, particularly with respect to complex effects Endothelial cells are grown on one side of the membrane, in the natural CNS environment. However, they are expen- and astrocytes are grown on the other, creating a simple bar- sive and time consuming, face ethical concerns, require a rier similar to the natural BBB environment. This basic 2 Journal of Healthcare Engineering model setup provides an approximation of the native BBB with barrier integrity and overall cell function. Here, we microenvironment but has several significant shortcomings describe the analysis of a 27-gene panel, consisting of genes compared to in vivo models. Membrane material, cell-cell associated with barrier integrity and endothelial cell or astro- contact, and fluid shear stress are some of the factors poorly cyte function. Prior to the gene expression panel, the suitabil- addressed by the standard transwell model. Two of these ity of multiple genes to serve as reference genes was examined issues, membrane material and cell-cell contact, can be to ensure an accurate comparison between the models. addressed simultaneously by changing the transwell insert membrane material. 2. Experimental Procedure Transwell insert membranes are a necessary artifact of barrier culture models, where a substrate is essential for cell 2.1. Cell Culture. Primary human astrocytes (HA) and cul- seeding, attachment, and monolayer formation. Synthetic ture media were obtained from ScienCell Research Laborato- barriers, such as polycarbonate or PET, are not readily ries (Carlsbad, CA, USA) and maintained according to the degraded or remodeled by either endothelial cells or astro- manufacturer’s instructions. Primary human brain micro- cytes, making it an unyielding barrier to “normal” BBB archi- vascular endothelial cells (HBMEC) and culture media were tecture. Without degradation, the membrane also imposes a obtained from Cell Systems (Kirkland, WA, USA) and main- hard limit on the degree of endothelial-to-astrocyte cell- tained according to the manufacturer’s instructions. Media cell contact. It has been repeatedly demonstrated that for both cell types contained 10% FBS. Cells used in all exper- proper gene expression and regulation of BBB proteins iments were at passage 1 when seeded. require direct contact between endothelial cells and astro- cytes [4, 5]. Additionally, a monolithic polymer barrier 2.2. Biopaper Insert Fabrication. Cell culture inserts with a between the cell types prevents proper extracellular matrix 15% electrospun gelatin biopaper membrane were fabricated (ECM) deposition and basement membrane formation, a fac- as previously described [8]. Briefly, 10 mL of a gelatin- tor known to affect BBB permeability [6, 7]. Taken together, genipin solution was prepared by dissolving 1.8 grams of type these conclusions point to the need for a biodegradable mem- B gelatin from bovine skin (Sigma-Aldrich, Saint Louis, MO, brane for in vitro models capable of supporting cell growth USA) into 70% (vol%) acetic acid (Sigma-Aldrich) in distilled and remodeling. water. The gelatin solution was magnetically stirred for 60 Previous research replaced the transwell PET membrane minutes before a solution consisting of 0.05 grams of genipin with an electrospun gelatin scaffold, styled “biopaper,” in an (Wako Chemicals, Richmond, VA, USA) dissolved in 0.5 mL endothelial/astrocyte coculture in vitro model [8]. This 21- of ethanol and 1 mL of 1X phosphate-buffered saline (PBS) day study demonstrated that the biopaper coculture models was added to the solution. The homogenous solution was were initially more permeable than standard PET models, then transferred to a 20 mL plastic syringe (Thermo Fisher at least until day 14. At later time points, the biopaper model Scientific, Rockwood, TN, USA) with a 22-gauge stainless was less permeable than the PET, as proven by both transen- steel blunt-ended needle (Jensen Global, Santa Barbara, CA, dothelial electrical resistance (TEER) measurements and USA). The spinning solution was stored between 24 and 72 dextran diffusion across the barrier. Notably, cell morphol- hours until electrospinning. ogy was similar in both models, indicating that a deeper look Gelatin nanofiber biopapers were electrospun using a into the gene expression profiles may be necessary to deter- Matsusada Precision AC100-240V high-voltage power sup- mine differences in transcript expression of essential BBB ply (Kusatsu city, Shiga, Japan) and a New Era Pump NE- proteins between the two models. 300 syringe pump (Farmingdale, NY, USA). The power To date, cellular gene expression profiling of in vitro BBB supply electrode was connected to the stainless steel models has been limited to only a handful of genes relevant to blunt-ended needle and positioned 15 cm from a grounded barrier permeability. Recent coculture models typically only circular stainless steel plate (10 cm diameter) covered with assay 5–10 genes, primarily focusing on endothelial cell junc- nonstick aluminum foil. The syringe pump was set to a tion proteins, with studies lasting for several days [9–11]. flow rate of 5 μL/min, and a voltage of 15 kV was applied Furthermore, several recent studies have employed the use to the electrode for a total of 30 minutes to produce each of immortalized cell lines or nonhuman animal cell lines. electrospun mat, designated “biopapers.” Electrospun bio- While this data is important for improving BBB models, gene papers were placed in a Sanplatec Dry Keeper desiccator expression profiles obtained from primary human cell cocul- (Kita-ku, Osaka, Japan) for at least 24 hours. ture models would be invaluable. Furthermore, an extensive Biopapers were cut to size and placed over the opening of comparison of BBB gene expression between models would hanging cell culture inserts (Millipore, Billerica, MA, USA) further illuminate the strengths and weaknesses of each from which the factory PET membrane was removed. The model system. edges of the biopaper were heat-sealed to the cell culture In order to further characterize the similarities and differ- insert. The excess biopaper around the edges was removed. ences between the biopaper and PET models, we sought to To cross-link the biopapers, the inserts were placed in determine differences in gene expression of BBB-related pro- 24-well plates, submerged in a 0.5 mL solution consisting teins at various time points. We hypothesize that the cell-cell of 5% (w/v) genipin dissolved in ethanol, and incubated for ° contact and the ability to remodel tissue scaffolding will 7 days at 37 C. Inserts were rinsed four times with sterile ° result in differential gene expression of functionally relevant water and stored in water at 4 C until needed, but not longer proteins at the BBB interface, including proteins associated than 7 days. As demonstrated in previous work [8], this Journal of Healthcare Engineering 3 protocol produces biopapers with an overall thickness of BSA and 0.1% Tween 20. Three final washes with PBS were 4.5 ± 1.5 μm. Prior to use, inserts were sterilized by placing performed on the samples, after which the Vectashield anti- under UV light for at least 20 minutes. fade mounting reagent was applied (H-1400, Vector Labs, Burlingame, CA) and coverslips were mounted. Samples 2.3. In Vitro BBB Model. In vitro bilayer human blood-brain were imaged the next day on a Nikon Eclipse Ti confocal barrier (BBB) models were prepared both with standard PET microscope using NIS-Elements software (NIS AR 4.10). cell culture inserts (1 μm pore size) and with electrospun bio- paper membranes. First, inserts were inverted and placed 2.5. Reverse Transcription Quantitative Polymerase Chain onto the cover of a 24-well plate. Fifty microliters of HA Reaction (RT-qPCR). Prior to RNA extraction, the samples medium containing 25,000 HAs was added to the bottom were centrifuged at 3200 rpm for 5 minutes and washed twice side of each insert. The well plate was used to cover the with 0.5 mL 1X PBS. Membranes were removed from the ° inserts which were placed in an incubator at 37 C for 2 hours original 1.5 mL collection tubes with sterile forceps and to allow for cell attachment. After 2 hours, the plates were transferred to sterile 2 mL tubes containing 350 mL Buffer righted and 1 mL of HA medium was added to the bottom RLT (Qiagen, Germantown, MD). A 5 mm stainless steel of each well. The following day, 500 μL of HBMEC medium bead (Qiagen, Germantown, MD) was added to the tube containing 25,000 HBMECs was added to the top of the cell followed by lysis on a Tissuelyser LT (Qiagen, Germantown, culture insert. Only the first passage of cells was used to min- MD) at 50 Hz for 2 minutes. Samples were spun at imize phenotypic changes from experiment to experiment. 15,000 rpm for 3 minutes, and the supernatant was trans- Medium formulation used for the experiment was identical ferred to a new 2 mL tube. Total RNA was purified with the to that used for cell culture prior to seeding. The medium RNeasy Plus Mini Kit (Qiagen, Germantown, MD) using was changed every 2-3 days. the QIAcube (Qiagen, Germantown, MD), an automated Inserts were harvested to analyze transcriptional differ- sample processing technology, according to the manufac- ences via reverse transcription quantitative polymerase chain turer’s instructions. The DNase digestion was extended to 1 reaction (RT-qPCR) to compare models between cells grown hour. Total RNA concentration and purity were quantified on the PET or biopaper. Using forceps, each membrane was using a Nanodrop 2000 spectrophotometer (Thermo Fisher pulled from the hanging plastic insert and fully submerged in Scientific, Waltham, MA). Total RNA was converted to com- the Allprotect Tissue Reagent (Qiagen, Germantown, MD) in plementary DNA (cDNA) using the RT2 First Strand Kit a 1.5 mL tube. Inserts were harvested after 3, 7, 14, 21, and 28 (Qiagen, Germantown, MD) protocol with 45 ng total RNA days with six inserts collected for both PET and biopaper per reaction. Each genomic DNA elimination mix contained membranes at each time point. Each biopaper and PET insert 45 ng total RNA and 2 μLBuffer GE and was brought to a was seeded with a different cell preparation, resulting in n =6 final volume of 10 μL with RNase-free water. Reactions were ° biological replicates. Samples in Allprotect were kept frozen incubated for 5 min at 42 C, whereupon they were placed on ° at −20 C until processed for RT-qPCR. ice and the components of the reverse transcription reaction added according to the manufacturer’s instructions, in a total ° 2.4. Fluorescent Immunohistochemistry. At day 21, three volume of 10 μL. Reactions were incubated at 42 C for ° samples from both the biopaper model and PET model were 15 min, followed by 5 min at 95 C. A total of 91 μL RNase- set aside for fluorescent immunohistochemistry analysis. free water was added to each 20 μL cDNA reaction and ° Prior to staining, all samples were removed from the trans- stored at −20 C until further use. It should be noted that well insert and placed on a glass slide. Samples were fixed extracted RNA from each sample continues the contribu- in 4% paraformaldehyde for 10 minute at room temperature, tions from both endothelial cells and astrocytes. Due to the followed by three five-minute washes with phosphate- experimental setup, it was impossible to separate endothelial buffered saline (PBS). Following fixation, all samples were cell RNA and astrocyte RNA. permeabilized with a 0.25% Triton X-100 PBS solution for 10 minutes at room temperature. Samples were washed three 2.6. Housekeeping Gene Arrays. A human housekeeping pro- times with PBS and incubated with a blocking solution of 1% filer RT2 PCR Array was used to assess a suite of genes for use bovine serum albumin (BSA) and 22 mg/mL glycine in PBS as potential reference genes. This array consisted of 12 house- containing 0.1% Tween 20 for 30 minutes at room tempera- keeping genes in a 96-well format (PAHS-000Z, Qiagen, ture. After blocking, samples were incubated with primary Germantown, MD). The stability of the genes was assessed antibodies for platelet adhesion cellular adhesion molecule on 28 samples representative of both biopaper and monocul- 1 (PECAM-1, mouse, Thermo Fisher 37-0700) and glial tures of all cell types. Each gene was assessed with two differ- fibrillary acidic (GFAP, chicken, Thermo Fisher ent final template (cDNA) concentrations: 1.6 ng and 0.1 ng. PA1-10004) in PBS containing 0.1% Tween 20 for 1 hour at Due to very low total RNA yields of monocultures, reference room temperature. Dilution of both primary antibodies was gene assays were not performed with these samples. The RT2 1 : 50. Following primary antibody incubation, samples were SYBR Green Master Mix (Qiagen, Germantown, MD) was washed with PBS three times, five minutes each, and incu- used for all reference gene assays. For all assays, each reaction bated with a secondary antibody solution for 1 hour at room consisted of 12.5 μL2xRT2 SYBR Green Mastermix, 2 μL temperature in the dark. The secondary antibody solution cDNA, and 10.5 μL nuclease-free water. The following proto- consisted of goat anti-mouse (SAB 4600082, Sigma) and goat col was used for all assays: an initial 10-minute incubation at ° ° ° anti-chicken (SAB 4600465, Sigma) in PBS containing 1% 95 C followed by 40 cycles of 95 C for 15 seconds and 60 C 4 Journal of Healthcare Engineering for 1 minute. A single reaction was performed for each gene Table 1: Housekeeping gene PCR efficiency and stability. on each sample. Reactions were recorded and analyzed using the Applied Biosystems ViiA™ 7 System software. The PCR Target gene PCR E 0.1 ng stability (SD) 1.6 ng stability (SD) efficiency of each gene was assessed via LinRegPCR [12]. ACTB 1.91 5.70 1.14 The expression stability of all candidate reference genes B2M 1.90 2.54 1.08 across all samples was assessed using the BestKeeper v. 1 soft- GUSB 1.93 0.84 ware program. [13] GAPDH 1.80 2.25 HPRT1 1.94 1.43 1.23 2.7. Custom RT2 PCR Array for RT-qPCR. A custom profiler RT2 PCR Array Format C was designed with 27 RT2 Primer HSP90AB1 1.95 1.46 1.02 Assays, 2 housekeeping genes (NONO and RPLP0), and 3 LDHA 1.94 1.84 controls (human genomic DNA contamination, reverse tran- NONO 1.96 1.39 0.84 scription, and positive PCR control) (SABiosciences). The 32 PGK1 1.92 4.11 1.21 genes were arranged in triplicate within a 96-well PCR array. PP1H 1.93 0.77 fl The array was prepared using the published protocol. Brie y, RPLP0 1.92 4.30 1.02 for one sample with analysis on 32 targets, 450 μL2xRT2 TFRC 1.91 3.12 1.06 SYBR Green Mastermix was added to 34 μL cDNA synthesis reaction and 416 μL RNase-free water and mixed well. Twenty-five microliters of the PCR component mix was types, though the stability of all genes was above the recom- added to each well of the custom array, and the plate was mended cut-off value of 1.0 (Table 1). At the high (1.6 ng) sealed with MicroAmp Optical Adhesive Film. After a brief cDNA concentration, the SD of all genes decreased, indi- spin, the array was placed in a Fast 96-well block of a ViiA cating greater stability. As with the low template concen- 7 Real-Time PCR System (Applied Biosystems, Foster City, tration, PP1H (SD = 0.77), NONO (SD = 0.84), and GUSB CA). The cycling conditions were an initial 10-minute (SD = 0.84) were the most stable of all genes tested, followed ° ° incubation at 95 C followed by 40 cycles of 95 C for 15 by RPLP0 and HSP90AB1 (both with SD = 1.02). NONO ° seconds and 60 C for 1 minute. The Applied Biosystems was chosen as the housekeeping gene for the custom ViiA 7 software was used to record and analyze all reactions. RT2 qPCR array, due to having high stability at both high The fluorescence threshold was manually adjusted for each and low concentrations. gene assay based on visual inspection of fluorescence in the log phase. 3.2. Fluorescent Immunohistochemistry. At day 21, fluores- cent staining clearly showed the presence of the BMEC 2.8. Data Analysis. The expression level of each gene within a marker PECAM-1 and the astrocyte marker GFAP given time point was compared between the biopaper and (Figure 1). The cobblestone pattern typical of an endothe- Δ PET coculture models using the CT method [14]. All sam- lial monolayer can also be readily observed, as well as an ples used NONO as a reference gene, having been validated elongated astrocyte morphology. Morphology of BMEC as the best housekeeping gene of the ones tested. Briefly, and astrocyte grown on the biopaper or PET is very similar, Δ the CT value of each gene was obtained by subtracting the indicating that the membrane materials may not have had a ff CT value of the reference gene (NONO) from the target gene. discernable e ect on cellular morphology. This value was linearized through a log2 transformation (i.e., −Δ −Δ 2 CT). Within each time point, the final 2 CT expression 3.3. Gene Array Data. Little to no expression was values for the biopaper and PET models were compared observed at any time point in either the biopaper or PET using two-tailed Student’s t-test, with results considered sig- coculture systems for genes β4 (GJB4), gap nificant at p <005. This procedure was repeated for all genes junction β6 (GJB6), tight junction protein 3 (TJP3), and clau- at all time points. Notably, two biopaper replicates and two din 3 (CLDN3) (Table 2). As these genes had no expression PET replicates from the day 7 time point were removed from values, statistical analysis could not be performed. all data analyses, as well as one PET replicate from day 3. At day 3, a significant difference in gene expression These samples were removed due to either complete lack of between coculture systems was observed for six genes expression of GFAP and AQP4 (known astrocyte-specific (Table 2). Three of these six genes had higher expression in fi markers) or signi cantly increased CT values (Table S1) biopaper cocultures, including gamma-glutamyltransferase [15, 16]. It is our belief that the astrocytes in these samples 5 (GGT5, p =0002), (TJP2, p < died and such they are not representative of the astrocyte/ 0 001), and cadherin-5 (CDH5, p =002). Conversely, cells endothelial model of this experiment. grown on PET had significantly higher expression for lami- nin subunit α1 (LAMA1, p =002), platelet-derived growth 3. Results factor receptor β (PDGFRB, p <0001), and leukemia inhib- itory factor (LIF, p =0018) at this time point. 3.1. Validation of Housekeeping Genes. All tested reference Significant differences in gene expression were observed genes possessed similar PCR efficiencies (Table 1). At the for nine genes at day 7. Cocultures grown on the biopaper low (0.1 ng) cDNA concentration, NONO was the most sta- had significantly higher expression of laminin subunit ble gene (standard deviation (SD) = 1.39) across all sample α4 (LAMA4, p =0018), platelet-derived growth factor β Journal of Healthcare Engineering 5

(a) (b) (c)

Electrospun biopaper membrane

BMEC (CD31) Astrocytes (GFAP) Merged (d) (e) (f)

PET membrane

Figure 1: Fluorescent immunohistochemistry of biopaper and PET membranes. PECAM-1 (red) is present at BMEC junctions in both models (a, d). GFAP (green) is distributed throughout the astrocyte, showing similar morphology across membrane materials (b, e). Merged images are shown in (c) and (f).

(PDGFB, p <0001), GGT5 (p =0004), CDH5 (p <0001), time points except day 14 (Figures 2(a) and 2(b), resp.). TJP2 (p =0001), and platelet endothelial cell adhesion PDGFB had significantly higher expression for all time molecule (PECAM1, p <0001). Compared to PET cocul- points after day 3 (Figure 2(c)), and VCAM1 was more tures, significantly lower expression was observed for highly expressed in the PET coculture model for days 14 secreted protein that is acidic and rich in cysteine (SPARC, and 21 only (Figure 2(d)). p =0003), collagen IV α1 (COL4A1, p =0003), and PDGFRB (p =0004). 4. Discussion Unlike previous time points, most of the eleven genes with differential expression at day 14 had higher expression In order to accurately model the BBB in vitro, cellular gene in PET cocultures compared to the biopaper cocultures. Spe- expression must be limited to only genes normally expressed cifically, PET cultures were observed to have significantly at the BBB in vivo. In order to confirm this for both the bio- higher expression of SPARC (p =0022), laminin subunit paper and PET models, GJB4 and GJB6 were included on the α2 (LAMA2, p =0006), LIF (p =0045), fibroblast growth array as negative controls. GJB4 and GJB6 code for gap junc- factor 2 (FGF2, p =0029), claudin 12 (CLDN12, p =0003), tion proteins 30.3 and connexin 30, respectively. tight junction protein 1 (TJP1, p =0008), gap junction α1 Connexin 30.3 expression has been demonstrated in non- protein (GJA1, p =0049), and vascular cell adhesion mole- BBB endothelial tissue, but there is currently no record of it cule 1 (VCAM1, p <0001). Only PDGFB (p =0004), TJP2 being expressed by either astrocytes or brain microvascular (p =003), and CDH5 (p =0001) had higher expression on endothelial cells. Thus, the absent expression of GJB4 in this the biopaper at this time point. study is consistent with literature findings, confirming that Day 21 had three genes with statistically significant neither model expresses barrier proteins not present at the expression differences between the two coculture systems. BBB. Unlike connexin 30.3, connexin 30 is a crucial compo- Cells grown on the biopaper had significantly higher expres- nent of gap junctions between astrocytes and plays an impor- sion of PDGFB (p =0007) and claudin 1 (CLDN1, p =0047) tant role in synaptic activity [17]. However, astrocytic but significantly lower expression of VCAM1 (p =0017)at expression is tightly controlled by neurons [18], which were this time point. not present in either the biopaper or PET models. As such, The final time point, day 28, had six genes with differen- the lack of GJB6 expression is readily explained and expected tial expression. Higher expression of LAMA4 (p =0021), by the lack of cellular signaling from neurons within the PDGFB (p <0001), PDGFRB (p =001), CDH5 (p <0001), culture. Taken together, these findings demonstrate that TJP2 (p =0035), and PECAM1 (p =0048) was observed in both coculture models do not display gene expression biopaper cocultures compared to PET cocultures. Notably, not associated with the BBB, due to either constitutive lack none of the assayed genes had higher expression in PET of expression or the requirement of additional cellular cues cultures at this time point. (i.e., neuron presence). Four genes, CDH5, TJP2, PDGFB, and VCAM1, with Expression of only endothelial, not epithelial, tight junc- significant roles in BBB maintenance and permeability had tion genes was confirmed though the use of negative controls significant differences in expression between models across TJP3 and CLDN3. TJP3, which codes for the tight junction multiple time points (Figure 2). CDH5 and TJP2 were signif- protein zonal occludens-3 (ZO-3), is primarily expressed at icantly more highly expressed in the biopaper model at all epithelial tight junctions [19]. To date, there is no evidence 6 Journal of Healthcare Engineering

Table 2: Significant differences in gene expression between the biopaper and PET coculture systems at each time point. Genes with significantly higher expression on the biopaper and PET are shown in italic and bold formats, respectively. Genes with no expression on the biopaper or PET at any time are presented with dashes.

Day 3 Day 7 Day 14 Day 21 Day 28 SPARC p = 0 003 p = 0 022 COL4A1 p = 0 003 LAMA1 p = 0 02 Extracellular matrix-associated proteins LAMA2 p = 0 006 LAMA4 p = 0 018 p = 0 018 AGRN GFAP Functional astrocyte proteins AQP4 PDGFB p < 0 001 p = 0 004 p = 0 007 p < 0 001 PDGFRB p < 0 001 p = 0 004 p = 0 01 Membrane transporters and soluble factors GGT5 p = 0 002 p = 0 004 LIF p = 0 018 p = 0 045 FGF2 p = 0 029 OCLN CLDN1 p = 0 047 Endothelial tight junction transmembrane proteins CLDN3 ————— CLDN5 CLDN12 p = 0 003 Endothelial adherens junction transmembrane proteins CDH5 p = 0 02 p < 0 001 p = 0 001 p < 0 001 TJP1 p = 0 008 Endothelial junction accessory proteins TJP2 p < 0 001 p = 0 001 p = 0 03 p < 0 001 TJP3 ————— GJA1 p = 0 049 Gap junction proteins GJB4 ————— GJB6 ————— VCAM1 p < 0 001 p = 0 02 Cell adhesion proteins PECAM1 p < 0 001 p = 0 048 of it being expressed at the BBB, consistent with findings in proteins. It also demonstrates for the first time that endothe- both the PET and biopaper models. Similarly, CLDN3 is lial cells grown on an electrospun biopaper model do not expressed at very low levels in brain tissue but instead has sig- have lower expression of these tight junction proteins than nificantly higher expression at epithelial barriers [20]. The those grown on traditional models. This is true for all assayed results obtained from these negative controls demonstrate endothelial transmembrane proteins except for claudin-12 that neither the biopaper nor PET model systems expressed (CLDN12) and claudin-1 (CLDN1). CLDN12, like CLDN5, proteins associated with epithelial, not endothelial, barriers. is located at tight junctions, though it is primarily associated Expression levels of all endothelial tight junction trans- with epithelial cells. To date, there is little record of CLDN12 membrane proteins assayed were equal between the biopaper expression in BBB-derived endothelial cells. Recent evidence and PET coculture models at all time points, with the excep- does suggest that CLDN12 is expressed by neural stem cells tion of CLDN12 and CLDN1. Interestingly, expression of and decreases upon differentiation [24]. The presence of (OCLN) and claudin-5 (CLDN5), the primary pro- quantifiable CLDN12 expression in both PET and biopaper teins comprising tight junctions [21], was not significantly model systems may be explained by some neural stem cells different between model systems. Low expression of occludin being present in the initial primary cell populations. This is a hallmark of several disrupted BBB disease states [22], and finding points to the limitation of this study; commercially claudin-5 plays a critical role in tight junction size-selective sourced primary cells are purified but may still have some permeability [23]. This finding suggests that the membrane residual cells capable of expansion and model alteration. This material, whether PET or gelatin biopaper, does not have a limitation is also supported by the commercial entities them- drastic effect on expression of high-abundance tight junction selves, as the endothelial cells are guaranteed to be >95% Journal of Healthcare Engineering 7

CDH5 ⁎ 1.20 ⁎ 0.60 TJP2

1.00 0.50 ⁎ ⁎ 0.80 ⁎ ⁎ 0.40 ⁎ ⁎ 0.60 0.30

0.40 0.20 Relative expression Relative Relative expression 0.20 0.10

0.00 0.00 Day 3 Day 7 Day 14 Day 21 Day 28 Day 3 Day 7 Day 14 Day 21 Day 28 BP BP PET PET (a) (b) PDGFB VCAM1 ⁎ 0.35 ⁎ ⁎ 8.00 0.30 0.25 6.00 ⁎ 0.20 ⁎ ⁎ 4.00 0.15 0.10 Relative expression Relative Relative expression 2.00 0.05 0.00 0.00 Day 3 Day 7 Day 14 Day 21 Day 28 Day 3 Day 7 Day 14 Day 21 Day 28 BP BP PET PET (c) (d)

Figure 2: Differences in gene expression across time points between the biopaper and PET for (a) CDH5, (b), TJP2, (c) PDGFB, and (d) VCAM1. In all graphs, gene expression is blue for the biopaper model and orange for the PET model. ∗p <005. pure, not necessarily 100%. Claudin-1 (CLDN1) forms the cell survival, stabilizing blood vessel assembly, and alter- tight junction barrier in conjunction with CLDN3, CLDN5, ing vascular permeability [26]. Low levels of CDH5 and OCLN and was more highly expressed in the biopaper expression are indicative of increased barrier permeability, model at one time point. Though there is some evidence that a method exploited by some viruses [27]. The higher increased expression of CLDN1 reduces BBB permeability to expression levels of CDH5 observed in the biopaper blood-borne molecules [25], it is probable that the observed model may signify more or tighter adherens junctions than level of expression did not have a quantifiable effect on bar- those observed in the PET model, indicative of higher barrier rier integrity differences between the models. The difference integrity or regulation. in expression, though significant, was slight (p =0047) and Despite the similarities in expression level of tight junc- did not persist for more than one time point. Additionally, tion transmembrane proteins between the biopaper and one PET sample was considerably lower than the rest, PET cocultures, the zonal occludens accessory protein family potentially causing the statistical difference observed. had different expression profiles. TJP1, coding for protein Future studies should incorporate more samples, in order zonal occludens-1 (ZO-1), had similar expression in the bio- to obtain datasets with lower standard deviation between paper and PET models at all time points except day 14, where samples. Overall, these findings indicate that extracellular the observed expression is significantly higher in the PET endothelial cell-cell adhesion and physical molecule imped- model. Formation of both tight and adherens junctions ance at tight junctions are most likely similar between the requires proper localization and assembly of ZO-1 scaffold- two coculture models. ing prior to recruitment of transmembrane proteins such as Higher expression of CDH5 at multiple time points occludin or claudins [28]. The similar level of expression indicates adherens junctions may result in decreased per- indicates that endothelial cells grown on biopaper or PET meability in the biopaper model. VE-cadherin, encoded most likely assemble tight and adherens junctions along sim- by CDH5, is the primary integral membrane protein of ilar timeframes. Interestingly, protein zonal occludens-2 endothelial adherens junctions [26]. It plays a multifunc- (ZO-2, expressed from TJP2) is significantly more highly tional role within the adherens junction, aiding in endothelial expressed in the biopaper model at all time points except 8 Journal of Healthcare Engineering day 21. Unlike ZO-1, ZO-2 is found exclusively at tight junc- On the other hand, the biopaper model may be more relevant tions, where it plays a significant role in junction assembly for investigating the progression of inflammation and disease [29]. Though not yet fully elucidated, it also appears to have states from a “healthy” baseline. a direct correlation with BBB integrity; multiple studies have It is particularly interesting that previous TEER testing shown a decrease in ZO-2 expression preceding increased does not correlate well with gene expression results obtained BBB permeability [30–32]. As such, the biopaper coculture in the current study. Prior analysis demonstrated that endo- model would be expected to be less permeable to molecular thelial/astrocyte bilayers grown on the biopaper had average diffusion at all time points, a finding not supported by previ- TEER values of 12 Ωcm2 at day 4 and 18 Ωcm2 at day 9, sig- ous research [8]. Future studies should look to clarify the dis- nificantly lower than the average TEER values of 14 Ωcm2 crepancy between mRNA levels and diffusion data by and 27 Ωcm2 recorded for PET bilayers at the same time assaying subcellular localization and abundance of ZO-2. points [8]. The expression data, particularly from barrier Secreted by endothelial cells [33], PDGFB plays a critical genes such as TJP2 and CDH5, would seem to indicate that role in BBB formation and maintenance, acting as a chemoat- the biopaper model should have a higher TEER value than tractant for pericytes [34] and altering gene expression. The the PET model at these early time points. However, there expression levels of PDGFB increased over time in the biopa- could be multiple reasons for the discrepancy between gene per model only, with levels remaining relatively constant in expression data and functional outcomes. One explanation the PET model. This pattern is reflected in expression levels is that, as previously noted, BBB permeability is a complex between the two models at each time point; no significant dif- process that requires many proteins working it in concert. ference was observed at the day 3 time point, but PDGFB was It is possible that higher gene expression of other accessory more highly expressed in the biopaper model at all later time proteins is required in order to achieve higher TEER levels points. Failure of endothelial cells to recruit pericytes results early. Another explanation is that there may be a delay in mechanical instability, eventually leading to microaneur- between gene expression and enough protein accumulation ysm [35] and embryonic lethality [36]. Though pericytes to effect functional changes. It is possible that a “threshold” were not present in either the PET or biopaper models, level of TJP2 or CDH5 is necessary to upregulate other cellu- higher PDGFB expression in the biopaper model may create lar changes leading to a tighter barrier, as is observed for a better gradient for putative pericyte chemoattraction. other genes [45]. Further research is needed to elucidate the Furthermore, the elevated expression level of PDGFB over timeline and correlation between BBB gene expression and time in the biopaper model indicates it would be better for a functional barrier permeability. long-term BBB model, as continued expression of PDGFB is necessary to maintain barrier integrity [37]. The addition of pericytes to both PET and biopaper models may further 5. Conclusions elucidate functional differences between models due to PDGFB expression. In vitro coculture models of the BBB comprised of endothe- Four genes were expressed at significantly higher levels at lial cells and astrocytes respond differently to transwell mem- multiple time points in the PET model. One of these genes, branes of either electrospun gelatin biopapers or PET. Of the vascular cell adhesion molecule 1 (VCAM1), is correlated 27 genes analyzed, 4 had little to no expression, 5 had no with increased BBB permeability. VCAM1 plays a critical significant difference in gene expression levels between the role in neutrophil migration into the CNS by opening pores biopaper and PET inserts, 10 had significantly higher expres- in the BBB, allowing cells to cross the barrier [38]. Endothe- sion in at least one time point in the PET model, and 7 had lial cells normally express VCAM1 at low levels but dramat- significantly higher expression in at least one time point for ically upregulate expression in response to inflammatory the biopaper model. The fact that less than a third of the conditions [39]. The increased expression and accompanying genes tested had similar expression between insert materials change in endothelial morphology are exploited by several means there was a significant difference between these two virus families for BBB penetration, as demonstrated both models. While both are valid models, our studies validate in vitro with dengue virus [40] and in vivo with Venezuelan the use of different membranes based on the experiments to equine encephalitis virus [41]. The result that VCAM1 was be tested. Both model systems have similar expression of significantly higher in the PET model compared to the biopa- integral membrane proteins at tight junctions but differ in per model at days 14 and 21 is relevant from a disease pathol- expression of accessory proteins such as ZO-2. The difference ogy standpoint, as it may be useful for inflammatory in expression of several key inflammatory genes such as modeling, further supporting the suggestion that the PET VCAM1 and LIF indicates that the PET model would be membrane is more indicative of an inflamed model. Like optimal for studying the effects of various therapies on BBB VCAM1, LIF is expressed at significantly higher levels in in the inflamed state. Conversely, the biopaper model has the inflamed central nervous system [42]. Though most higher expression of genes associated with a “tighter” BBB research on the effects of LIF on the CNS has been focused for culture times of at least 28 days, thus validating the use on the spinal cord, multiple studies have shown it to be a of the biopaper model for long-term studies. Future experi- potent proinflammatory cytokine [43, 44]. The significantly ments should focus on a subset of these tested genes and cor- higher levels present in the PET model suggest the model relate mRNA level with protein abundance. Furthermore, the may be more indicative of an “inflamed” state, potentially application of the PET model for studying the inflamed BBB useful for testing the effects of a treatment on inflammation. should be further investigated. Journal of Healthcare Engineering 9

Conflicts of Interest with high transendothelial electrical resistance,” Brain Research, vol. 1521, pp. 1–15, 2013. fl The authors declare that there is no con ict of interest [11] N. Perriere, S. Yousif, S. Cazaubon et al., “A functional in vitro regarding the publication of this article. All work done model of rat blood-brain barrier for molecular analysis of by Joel D. Gaston, Lauren L. Bischel, and Kathleen D. efflux transporters,” Brain Research, vol. 1150, pp. 1–13, 2007. Cusick was done while under contract with the U.S. Naval [12] C. Ramakers, J. M. Ruijter, R. H. L. Deprez, and A. F. M. Research Laboratory. Moorman, “Assumption-free analysis of quantitative real- time polymerase chain reaction (PCR) data,” Neuroscience Let- Acknowledgments ters, vol. 339, no. 1, pp. 62–66, 2003. [13] M. W. Pfaffl, A. Tichopad, C. Prgomet, and T. P. Neuvians, This work was supported by the Chemical Biological “Determination of stable housekeeping genes, differentially Technologies Directorate under Project no. CB10011 from regulated target genes and sample integrity: BestKeeper - the U.S. Department of Defense Chemical and Biological Excel-based tool using pair-wise correlations,” Biotechnology Defense Program through the Defense Threat Reduction Letters, vol. 26, no. 6, pp. 509–515, 2004. Agency (DTRA), the American Society for Engineering [14] K. J. Livak and T. D. Schmittgen, “Analysis of relative gene Education (ASEE), and the National Research Council expression data using real-time quantitative PCR and the −ΔΔCT (NRC). The authors would like to thank Jaimee Compton. 2 method,” Methods, vol. 25, no. 4, pp. 402–408, 2001. [15] T. Badaut, T. Lasbennes, P. J. Magistretti, and L. Regli, “Aqua- Supplementary Materials porins in brain: distribution, physiology, and pathophysiol- ogy,” Journal of Cerebral Blood Flow and Metabolism, vol. 22, no. 4, pp. 367–378, 2002. Absent gene expression in 4 samples for both AQP4 and “ fi GFAP. The highlighted samples were removed from all [16] L. F. Eng, R. S. Ghirnikar, and Y. L. Lee, Glial brillary acidic protein: GFAP-thirty-one years (1969-2000),” Neurochemical analyses (Supplementary files). Research, vol. 25, no. 9-10, pp. 1439–1451, 2000. [17] U. Pannasch, L. Vargova, J. Reingruber et al., “Astroglial References networks scale synaptic activity and plasticity,” Proceedings of the National Academy of Sciences of the United States of [1] A. Wolff, M. Antfolk, B. Brodin, and M. Tenje, “In vitro blood– America, vol. 108, no. 20, pp. 8467–8472, 2011. brain barrier models—an overview of established models and ff “ new microfluidic approaches,” Journal of Pharmaceutical Sci- [18] A. Koulako , P. Ezan, and C. Giaume, Neurons control the ences, vol. 104, no. 9, pp. 2727–2746, 2015. expression of connexin 30 and connexin 43 in mouse cortical astrocytes,” Glia, vol. 56, no. 12, pp. 1299–1311, 2008. [2] R. Booth and H. Kim, “Characterization of a microfluidic in vitro model of the blood-brain barrier (mu BBB),” Lab on [19] A. Inoko, M. Itoh, A. Tamura, M. Matsuda, M. Furuse, and “ a Chip, vol. 12, no. 10, pp. 1784–1792, 2012. S. Tsukita, Expression and distribution of ZO-3, a tight “ junction MAGUK protein, in mouse tissues,” Genes to Cells, [3] M. B. Esch, T. L. King, and M. L. Shuler, The role of body-on- – a-chip devices in drug and toxicity studies,” in Annual Review vol. 8, no. 11, pp. 837 845, 2003. “ of Biomedical Engineering, M. L. Yarmush, J. S. Duncan and M. [20] K. J. Hewitt, R. Agarwal, and P. J. Morin, The claudin gene ” L. Gray, Eds., vol. 13, , pp. 55–72, Palo alto, CA, USA, 2011. family: expression in normal and neoplastic tissues, BMC [4] N. J. Abbott, L. Ronnback, and E. Hansson, “Astrocyte-endo- Cancer, vol. 6, no. 1, 2006. thelial interactions at the blood-brain barrier,” Nature Reviews [21] E. Steed, M. S. Balda, and K. Matter, “Dynamics and functions Neuroscience, vol. 7, no. 1, pp. 41–53, 2006. of tight junctions,” Trends in Cell Biology, vol. 20, no. 3, – [5] A. A. Hurwitz, J. W. Berman, W. K. Rashbaum, and W. D. pp. 142 149, 2010. Lyman, “Human fetal astrocytes induce the expression of [22] B. T. Hawkins and T. P. Davis, “The blood-brain barrier/neu- blood-brain barrier specific proteins by autologous endothe- rovascular unit in health and disease,” Pharmacological lial-cells,” Brain Research, vol. 625, no. 2, pp. 238–243, 1993. Reviews, vol. 57, no. 2, pp. 173–185, 2005. [6] S. Liebner, C. J. Czupalla, and H. Wolburg, “Current concepts [23] T. Nitta, M. Hata, S. Gotoh et al., “Size-selective loosening of of blood-brain barrier development,” International Journal of the blood-brain barrier in claudin-5-deficient mice,” Journal Developmental Biology, vol. 55, no. 4-5, pp. 467–476, 2011. of Cell Biology, vol. 161, no. 3, pp. 653–660, 2003. [7] C. Hartmann, A. Zozulya, J. Wegener, and H. J. Galla, “The [24] A. K. Watters, S. Rom, J. D. Hill et al., “Identification and impact of glia-derived extracellular matrices on the barrier dynamic regulation of tight junction protein expression in function of cerebral endothelial cells: an in vitro study,” Exper- human neural stem cells,” Stem Cells and Development, imental Cell Research, vol. 313, no. 7, pp. 1318–1325, 2007. vol. 24, no. 12, pp. 1377–1389, 2015. [8] L. L. Bischel, P. N. Coneski, J. G. Lundin et al., “Electrospun [25] F. Pfeiffer, J. Schafer, R. Lyck et al., “Claudin-1 induced sealing gelatin biopapers as substrate for in vitro bilayer models of of blood-brain barrier tight junctions ameliorates chronic blood-brain barrier tissue,” Journal of Biomedical Materials experimental autoimmune encephalomyelitis,” Acta Neuro- Research Part A, vol. 104, no. 4, pp. 901–909, 2016. pathologica, vol. 122, no. 5, pp. 601–614, 2011. [9] G. L. Li, M. J. Simon, L. M. Cancel et al., “Permeability of [26] S. Tietz and B. Engelhardt, “Brain barriers: crosstalk between endothelial and astrocyte cocultures: in vitro blood-brain bar- complex tight junctions and adherens junctions,” Journal of rier models for drug delivery studies,” Annals of Biomedical Cell Biology, vol. 209, no. 4, pp. 493–506, 2015. Engineering, vol. 38, no. 8, pp. 2499–2511, 2010. [27] P. Shrivastava-Ranjan, P. E. Rollin, and C. F. Spiropoulou, [10] A. Patabendige, R. A. Skinner, and N. J. Abbott, “Establish- “Andes virus disrupts the endothelial cell barrier by induction ment of a simplified in vitro porcine blood-brain barrier model of vascular endothelial growth factor and downregulation of 10 Journal of Healthcare Engineering

VE-cadherin,” Journal of Virology, vol. 84, no. 21, pp. 11227– [42] K. S. Na, H. Y. Jung, and Y. K. Kim, “The role of 11234, 2010. pro-inflammatory cytokines in the neuroinflammation ” [28] A. S. Fanning and J. M. Anderson, “Zonula occludens-1 and -2 and neurogenesis of schizophrenia, Progress in Neuro- – are cytosolic scaffolds that regulate the assembly of cellular Psychopharmacology & Biological Psychiatry, vol. 48, pp. 277 junctions,” in Molecular Structure and Function of the Tight 286, 2014. Junction: From Basic Mechanisms to Clinical Manifestations, [43] B. J. Kerr and P. H. Patterson, “Potent pro-inflammatory M. Fromm and J. D. Schulzke, Eds., vol. 1165, Annals of the actions of leukemia inhibitory factor in the spinal cord of the New York Academy of Sciences, New York, NY, USA, 2009. adult mouse,” Experimental Neurology, vol. 188, no. 2, [29] U. Kniesel and H. Wolburg, “Tight junctions of the blood- pp. 391–407, 2004. brain barrier,” Cellular and Molecular Neurobiology, vol. 20, [44] I. Pineau and S. Lacroix, “Proinflammatory cytokine synthesis no. 1, pp. 57–76, 2000. in the injured mouse spinal cord: multiphasic expression pat- [30] N. Shimojima, C. B. Eckman, M. McKinney et al., “Altered tern and identification of the cell types involved,” Journal of expression of zonula occludens-2 precedes increased blood- Comparative Neurology, vol. 500, no. 2, pp. 267–285, 2007. brain barrier permeability in a murine model of fulminant [45] H. H. McAdam and A. Arkin, “Stochastic mechanisms in gene hepatic failure,” Journal of Investigative Surgery, vol. 21, expression,” Proceedings of the National Academy of Sciences, no. 3, pp. 101–108, 2008. vol. 94, no. 3, pp. 814–819, 1997. [31] S. W. Lee, W. J. Kim, H. O. Jun, Y. K. Choi, and K. W. Kim, “Angiopoietin-1 reduces vascular endothelial growth factor- induced brain endothelial permeability via upregulation of ZO-2,” International Journal of Molecular Medicine, vol. 2, no. 2, pp. 229–233, 2009. [32] I. E. Andras, H. Pu, M. A. Deli, A. Nath, B. Hennig, and M. Toborek, “HIV-1 Tat protein alters tight junction protein expression and distribution in cultured brain endothelial cells,” Journal of Neuroscience Research, vol. 74, no. 2, pp. 255–265, 2003. [33] K. Gaengel, G. Genove, A. Armulik, and C. Betsholtz, “Endothelial-mural cell signaling in vascular development and angiogenesis,” Arteriosclerosis Thrombosis and Vascular Biology, vol. 29, no. 5, pp. 630–638, 2009. [34] A. Armulik, G. Genove, and C. Betsholtz, “Pericytes: devel- opmental, physiological, and pathological perspectives, prob- lems, and promises,” Developmental Cell, vol. 21, no. 2, pp. 193–215, 2011. [35] P. Lindahl, B. R. Johansson, P. Leveen, and C. Betsholtz, “Peri- cyte loss and microaneurysm formation in PDGF-B-deficient mice,” Science, vol. 277, no. 5323, pp. 242–245, 1997. [36] E. A. Winkler, R. D. Bell, and B. V. Zlokovic, “Central nervous system pericytes in health and disease,” Nature Neuroscience, vol. 14, no. 11, pp. 1398–1405, 2011. [37] R. D. Bell, E. A. Winkler, A. P. Sagare et al., “Pericytes control key neurovascular functions and neuronal phenotype in the adult brain and during brain aging,” Neuron, vol. 68, no. 3, pp. 409–427, 2010. [38] S. van Wetering, N. van den Berk, J. D. van Buul et al., “VCAM-1-mediated Rac signaling controls endothelial cell- cell contacts and leukocyte transmigration,” American Journal of Physiology-Cell Physiology, vol. 285, no. 2, pp. C343–C352, 2003. [39] J. I. Alvarez, R. Cayrol, and A. Prat, “Disruption of central nervous system barriers in multiple sclerosis,” Biochimica Et Biophysica Acta-Molecular Basis of Disease, vol. 1812, no. 2, pp. 252–264, 2011. [40] M. L. Velandia-Romero, M. A. Calderon-Pelaez, and J. E. Castellanos, “In vitro infection with dengue virus induces changes in the structure and function of the mouse brain endothelium,” PLoS One, vol. 11, no. 6, 2016. [41] A. Sharma, M. Bhomia, S. P. Honnold, and R. K. Maheshwari, “Role of adhesion molecules and inflammation in Venezuelan equine encephalitis virus infected mouse brain,” Virology Journal, vol. 8, 2011. International Journal of Rotating Machinery

International Journal of Journal of The Scientific Journal of Distributed (QJLQHHULQJ World Journal Sensors Sensor Networks Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 201 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of Control Science and Engineering

Advances in Civil Engineering Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at https://www.hindawi.com

Journal of Journal of Electrical and Computer Robotics Engineering Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

VLSI Design Advances in OptoElectronics ,QWHUQDWLRQDO-RXUQDORI Modelling & International Journal of Simulation $HURVSDFH Navigation and in Engineering Observation (QJLQHHULQJ

Hindawi Publishing Corporation Hindawi Publishing Corporation +LQGDZL3XEOLVKLQJ&RUSRUDWLRQ Hindawi Publishing Corporation Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 KWWSZZZKLQGDZLFRP 9ROXPH http://www.hindawi.com Hindawi Publishing Corporation http://www.hindawi.com Volume 201-

International Journal of International Journal of Antennas and Active and Passive Advances in Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration

Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014